• by Josh Chasin , Op-Ed Contributor, December 14, 2007

Several weeks ago in this space, David Smith urged Randy Rothenberg (IAB) and Jim Sterne (WAA) to get together over the reach standards and definitions their two organizations are putting out. Following that column, Judah Phillips then wrote about the yin and yang of online metrics -- audience measurement and Web analytics -- from his perspective as a Web analytics guru. Judah noted that Web analytic data and audience measurement data can and should co-exist, that the two sciences are best used to address different sets of questions.

I recently had the privilege of participating in the morning panel discussion at the IAB's Audience Measurement Leadership Forum, moderated by MediaPost's own erstwhile correspondent, Joe Mandese. As comScore's Chief Research Officer and a career audience measurement guy (or, as Ad Age recently noted, a "quantifier par excellence") maybe I'm biased, but I thought we made some progress in disentangling the whole "panel-versus-server" morass. So let me lay out my take on that topic here. Consider this a companion piece to Judah's column, covering similar ground but from the audience measurement perspective.

Audience measurement data has evolved over about 80 years with the objective of providing both buyers and sellers of media with a common currency, or consensus look at audiences across media vehicles. This "look" includes demographic descriptors (traditionally at minimum age and gender) sufficient to allow for tracking audience behavior by media buying target. It also includes the ability to combine and de-dupe audiences across media vehicles in aggregate. These aggregates are called "schedules," so audience measurement data allows for the calculation of schedule metrics like unduplicated reach, frequency of exposure, and gross impressions (or GRPs.)

This is not to say audience measurement panels are perfect. There will always be vigorous dialogue around the numbers, always room for intelligent parties to disagree. (Or, as articulated at the client meeting I just got back from, "higher numbers are popular.") But at least the disagreements can commence from that common data set.

It is a truism of audience measurement that data facilitates commerce, and commerce is good. The first step in commerce is dialogue, negotiation.

Web analytics, on the other hand, is a science that has emerged around the purely digital nature of the Internet, and the fact that machines talking to machines leave trails, or breadcrumbs -- or heck, let's just call them cookies. Web analytic data is used to develop Key Performance Indicators (KPIs); to manage and optimize the Web site; to make site design and eCommerce decisions; and to communicate internally about performance. This latter point is an important one; site-centric data has been developed for internal purposes, it is not designed to be consensus or "show" data. It is certainly not designed for ad sales; I've never heard an agency ask if Omniture could interface with their buying systems.

When I joined comScore in May, one of the first things I did was to join the Web Analytics Association's listserv. You can check out recent posts or join the listserv here, and you should, especially if you are an online metrics professional. I'm sure you would agree that this is not a forum that has a whole lot to do with buying and selling advertising. It's a great forum, where a bunch of really smart people trade knowledge and help each other do their jobs better. But buying and selling advertising is simply not the domain of Web analytics professionals.

Now at this point, you're probably thinking, "OK, fine, these are two different animals. I get it. But which is right?" And my answer to that would be, yes.

Right now the IAB, in partnership with George Ivie and the MRC, is putting the finishing touches on a document entitled IAB Audience Reach Measurement Guidelines. This document will help to put the industry on the same page with respect to the characteristics of unique reach estimates as provided by panel-based audience measurement companies, site-centric Web analytics companies, and third-party ad servers. It establishes that audience measurement companies are endeavoring to identify and quantify Unique Visitors or Unique Users, which is to say, unduplicated persons. If you've worked in other media, you've probably called that "cume." The document also makes clear that Web analytics providers typically provide counts of Unique Cookies.

Cookies are bits of code. People are the wetware that actually look at the ads and buy the products. Measuring that code is neat and tidy, which has led some to contend that site-centric data is "census" and "perfect." And for the purposes of Web analytics, that may be so. But for those of us in audience measurement, getting at the behavior of those darned people out there is the essence of what we contribute to the value chain. And it is, I like to think, where much of the art lies.

Is there any hope of reconciliation? There is a path, if not a panacea. The IAB document (which, I should caution, I've only seen in draft), defines that path. Site-centric metrics providers can use the term Unique Users or Unique Visitors if they are able to account for certain phenomena, notably including cookie deletion. (Have I directed you to the comScore Cookie Deletion White Paper yet? Surely I must have.) They will also have to account for duplication across home and work, and to account for handling of non-human traffic.

I'll admit that I'm still learning about this business, and I hope that never stops. I just learned about another issue confronting site-centric metrics providers: tabbed browsers. If a user sets his or her browser to open each morning with five tabs, each representing a different publisher, the site-centric data provider for each of those publishers will log a visit and at least one page view when the browser opens. The person out there in front of the screen sees one web page, but five different publishers got server calls, and five different site-centric providers logged a visit. Because panel companies focus on the experience and engagement of persons, not machines, we will only attribute a visit to the active tab; and then, to subsequent tabs if and only if they become active. So now, on my slide comparing Unique Cookies to Unique Visitors, I've had to add a bullet: tabbed browsing.

I very much appreciated reading Judah's column, and I look forward to more hands-across-the-Internet bonding between those of us who come from the audience measurement school, and those of you who come from the Web analytics school. Our fields are different, but they overlap, and we certainly share a customer constituency. I think we can learn from each other. We could collaborate on studies and analyses that would be of great benefit to the industry. I'm looking forward to pushing that conversation forward in 2008.

Happy holidays to all, and don't forget to friend me on Facebook!